EP0936575A1 - Procédé de traitement d'images pour l'estimation de mouvement dans une séquence d'images, procédé de filtrage du bruit et appareil d'imagerie médicale utilisant ces procédés - Google Patents
Procédé de traitement d'images pour l'estimation de mouvement dans une séquence d'images, procédé de filtrage du bruit et appareil d'imagerie médicale utilisant ces procédés Download PDFInfo
- Publication number
- EP0936575A1 EP0936575A1 EP99200360A EP99200360A EP0936575A1 EP 0936575 A1 EP0936575 A1 EP 0936575A1 EP 99200360 A EP99200360 A EP 99200360A EP 99200360 A EP99200360 A EP 99200360A EP 0936575 A1 EP0936575 A1 EP 0936575A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- image
- causal
- block
- images
- criterion
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 70
- 238000003672 processing method Methods 0.000 title claims abstract description 12
- 238000001914 filtration Methods 0.000 title claims description 27
- 238000002059 diagnostic imaging Methods 0.000 title claims description 4
- 239000013598 vector Substances 0.000 claims abstract description 91
- 230000001364 causal effect Effects 0.000 claims abstract description 56
- 230000002123 temporal effect Effects 0.000 claims abstract description 33
- 238000012360 testing method Methods 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 10
- 238000011156 evaluation Methods 0.000 claims 1
- 230000004807 localization Effects 0.000 abstract 2
- 238000004364 calculation method Methods 0.000 description 10
- 230000003044 adaptive effect Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 238000000354 decomposition reaction Methods 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 230000000630 rising effect Effects 0.000 description 3
- 102100030386 Granzyme A Human genes 0.000 description 2
- 101001009599 Homo sapiens Granzyme A Proteins 0.000 description 2
- 230000009977 dual effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000002601 radiography Methods 0.000 description 2
- 101000614502 Caenorhabditis elegans Adenylate kinase isoenzyme 6 homolog Proteins 0.000 description 1
- 229920000297 Rayon Polymers 0.000 description 1
- 230000010339 dilation Effects 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 230000008571 general function Effects 0.000 description 1
- 230000009931 harmful effect Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 239000002964 rayon Substances 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 238000011524 similarity measure Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 230000002792 vascular Effects 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/223—Analysis of motion using block-matching
- G06T7/238—Analysis of motion using block-matching using non-full search, e.g. three-step search
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
Definitions
- the invention relates to an image processing method for estimating the movement of objects represented in a sequence of images noise, as well as a noise filtering method including this method of estimating movement.
- the invention also relates to a medical imaging device. implementing the motion estimation method and / or the method of noise filtering, and in particular an X-ray machine.
- this X-ray machine works in fluoroscopic mode for the visualization of the vascular system.
- a block of same coordinates is determined and a search window whose dimensions are related to the assumed range of motion is formed around this block.
- the purpose of this process based on an algorithm called B.M.A is to determine in the search window a block called block corresponding to the current block which is the one which, according to a criterion intensity similarity preset, has the closest brightness properties those of the current block.
- the determination of the corresponding block provides a vector of motion which measures the spatial translation between the current block in the image at time t and the corresponding block in the previous image at time t-1.
- This motion estimation method includes an estimation of a coarse motion vector to provide an approximate location of the block corresponding, followed by the application of the intensity similarity criterion to estimate a real motion vector around the coarse motion vector.
- the criterion similarity is not described.
- the rough motion vector estimate includes a determination of a plurality of p motion vectors relating to blocks correspondents located respectively in an identical number p of images successive temporal periods prior to the image processed at time t which contains the block current, and a linear temporal filtering which provides said motion vector rough as a weighted average of the p vectors of previous movements.
- This linear temporal filtering of the p prior motion vectors is of the type predictive, that is to say that knowing p vectors of relative anterior movements with p corresponding blocks taken from previous images, a prediction of the coarse motion vector between instants t and t-1 is realized based on a specific hypothesis that the estimated motion on the anterior p blocks successive correspondents is strictly linear and continuous. The criterion of similarity is then applied locally around the coarse vector to estimate the vector of real movement.
- this process involves this prediction step, it loses its efficiency whenever a new movement appears, because, in this case, the coarse motion vector cannot be determined since the p vectors of previous movements do not exist. And so the real vector based on the vector coarse also cannot be found locally.
- An object of the invention is to provide a method for estimating movement serving in particular to assist a temporal filter, or spatio-temporal for effective denoising in the event of movement.
- An image processing method for noise reduction in a image sequence includes an arrangement of recursive time filtering and such a method for motion estimation according to claim 8.
- a medical imaging device includes an acquisition system of a sequence of images, and an image processing system that has access to image data and a sequence image display system which includes a processor implementing such an image processing method for noise reduction.
- the invention relates to a method for processing a sequence of images.
- noises representing objects having movements that take place on a substantially large area relative to the size of an image, or large amplitude from one image to another.
- large amplitude we mean movements of the order of or greater than 10 pixels.
- One area of application may be noise reduction in video images, especially images that need to be encoded because the encoding of noisy footage is expensive.
- Another area of application is that of reduction noise in medical images, especially images provided by X-ray radiology equipment working in fluoroscopic mode.
- These devices provide, at rates of the order of 3 to 25 frames per second, sequences very noisy images because they are formed using very low doses of X-rays.
- These sequences of fluoroscopic images are frequently used to follow live operating gestures and therefore must be denoised in time real.
- these image sequences include very fine objects in movement, such as a catheter for example, which should not be cleared by the image processing. By fine details or fine objects we mean objects of diameter on the order of or less than 10 pixels.
- the motion estimation method comprises a step 10 of acquisition and digitization of a sequence of images comprising an image to be processed denoted J t , a previous image called causal denoted J t-1 and a following image called anti- causal noted J t + 1 .
- each digitized noisy image J t , J t-1 , J t + 1 is a two-dimensional spatial matrix of pixels identified by coordinates x, y, which are denoted respectively A t , A t- 1 , A t + 1 in the three temporal images and which respectively have a digitized intensity level I t (x, y), I t-1 (x, y), I t + 1 (x, y).
- the intensities are directly indicated by I t , I t-1 , I t + 1 .
- a method of recursive temporal noise filtering capable of reducing the noise of each image successively in a sequence of images comprises a processing of a signal called noisy temporal signal formed from the different levels of intensity I t-1 , I t , I t + 1 of the pixels A t-1 , A t , A t + 1 located at x, y as a function of time t (tau).
- the time signal shows a rising intensity front between the instants t-1 and t, with a plateau beyond t, at time t + 1, which is characteristic of a movement between instants t-1 and t + 1, in an otherwise simply noisy signal.
- the time signal shows a noise peak at time t, between the instants t-1 and t + 1, therefore a signal which is simply entirely noisy.
- a temporal filtering method was applied to the signal of FIG. 7A to obtain a filtered temporal signal denoted P (t) by a recursive temporal filtering without compensation for movement.
- P (t) a filtered temporal signal denoted by a recursive temporal filtering without compensation for movement.
- the intensities remain offset on either side of the rising edge due to the movement.
- FIG. 7D shows the same filtered temporal signal of the noise in which the rising edge no longer exists, which means that the movement is also compensated.
- the present method makes it possible to estimate the movement in order to compensate for it in order to obtain the result of FIG. 7D.
- the temporal intensities I t-1 , I t , I t + 1 which are the points of the temporal signal are called samples.
- the samples preceding and following the treated sample I t are called causal and anti-causal respectively and denoted I t-1 and I t + 1 .
- Recursive temporal filtering constitutes according to the knowledge of one of the best methods for denoising images. However this type of filtering is not able to filter the image correctly in the event that a movement of large amplitude appears.
- a solution to nevertheless use the recursive temporal filtering in the event of movement is to apply a method of image processing which includes a first phase of motion estimation and a second phase of recursive time filtering. The combination of these two phases can be done according to different schemes.
- the noisy signal It of the image J t to be processed is subjected to a first phase 150 in which both an estimation and a compensation of the movements relating to the processed image J t .
- This first phase provides a compensated signal C t (English: Registered).
- This compensated signal C t is then subjected to a second phase 250 which performs the recursive time filtering and supplies the denoised signal P t .
- the noisy signal I t of the image J t to be processed is subjected to a first phase 100 which performs only a motion estimation and which provides a relative motion vector V t at location x, y.
- the noisy signal I t and the motion vector V t are then subjected to a second phase 200 which performs the temporal filtering of the signal I t aided by the motion vector V t to compensate for the motion and produce the denoised signal P t to be assigned at the current pixel to build a denoised image.
- the time filter receives both the intensities I t of the images and a motion vector field V t , and uses the motion vector field to compensate for the motion. in the image at the same time as it filters it.
- An object of the invention is therefore to provide a method for estimating movement to carry out the first phase of the diagrams presented above.
- Motion estimation is a process in itself used to denoise images to be combined with a second phase of time filtering, for a effective denoising in the event of movement.
- Many methods of estimating movement already exists in the state of the art, but these methods generally apply to low noise images, so that when they are applied to heavily noisy images the results are catastrophic.
- the present motion estimation method comprises a preliminary step 20 where the image to be processed J t is cut into contiguous blocks of dimension 12 x 12 or 16 x 16 pixels for example.
- the image J t is thus paved in elementary blocks called reference blocks denoted RBt (English: Referenced Block).
- the reference blocks are processed one after the other according to a fully automatic scanning method known to those skilled in the art, for example from left to right and from top to bottom of the image.
- a current block which is a reference block chosen at a given position in the image to be processed J t .
- Each reference block of the image J t is defined by a reference point Mk which is located for example in the center of the block.
- the reference point is always considered in the same place in each block and is characterized by its x, y coordinates, as in FIG. 4.
- the reference block itself constitutes a neighborhood noted NG (x, y) of the reference point M k .
- the present motion estimation method denoted FA-BMA comprises a matching of three blocks respectively in three temporal images constituted by the image to be processed J t , the causal image J t-1 and the anti image -causal J t + 1
- the motion estimation algorithm already known from the prior art denoted BMA (from the English Block Matching Algorithm) performs the mapping of only two blocks in two images respectively which are only the image to be processed and the causal image.
- the research policy stage therefore uses not only two successive images as already known but a triplet of images consisting of the causal image J t-1 , the image to be processed J t and an additional image which is the anti-causal image J t + 1 .
- the step 30 of applying the specific search policy SP comprises on the one hand, for a given reference block RB t a substep 31 for determining a homologous block denoted HB t-1 in the causal image J t-1 which has the same location x, y as the reference block RB t and around the homologous block HB t-1 a definition of a search window denoted SW t- 1 ( English: Search Window) in which we define a certain number of blocks called candidates.
- SW t- 1 English: Search Window
- Step 30 of applying the specific search policy SP further comprises, for the same reference block RB t , a sub-step 32 of determining an homologous anti-causal block HB t + 1 in l anti-causal image J t + 1 , defining a search window SW t + 1 around the homologous anti-causal block in which candidate blocks are defined according to the same process.
- search windows SW t-1 , SW t-1 by their size, which defines the maximum modulus of the motion vectors that can be found between the instants t and t-1 on the one hand, and t and t + 1 on the other hand.
- the vector stops at the limits of the search window which is therefore determined according to the maximum movements that are expected in the image, or that one wants to compensate.
- Step 40, 42 of applying the double intensity similarity criterion DSC comprises on the one hand, a comparison of the candidate blocks of the causal image J t-1 with the reference block RB t , and comprises on the other hand share a comparison of the candidate blocks of the anti-causal image J t + 1 to the same reference block RB t with the same double criterion of similarity of intensity applied to this anti-causal image instead of the causal image.
- the candidate blocks which most resemble the reference block RB t according to this double criterion of intensity similarity are chosen as corresponding blocks and are denoted respectively CB t-1 and CB t + 1 to define a first motion vector denoted V ( t, t-1) between time t and time t-1 at location x, y. and to define a second motion vector denoted V (t, t + 1) between the instants t and t + 1 at the same location x, y.
- the first motion vector V (t, t-1) spatially connects the center Mk t of the reference block RB t to the center Mk t-1 of the corresponding block CB t-1 and is oriented from the reference block RB t in the image J t towards the homologous block HB t-1
- the second motion vector V (t, t + 1) spatially connects the center Mk t of the reference block RBt to the center Mk t + 1 of the corresponding block CB t + 1 and is oriented from the reference block RBt in the image J t towards the homologous anti-causal block HB t + 1 .
- the new double similarity criterion D-SC is based on an assumption that the movement between the corresponding block in image t-1 and the reference block in image t on the one hand, and between the reference block in image t and the corresponding block in image t + 1 is locally linear, continuous and uniform. That is, the motion vector between t-1 and t is the same as between t and t + 1.
- the vector V (t-1, t) is the same as the vector V (t, t + 1), has the same module and the same orientation.
- an elementary similarity criterion which forms a first part called causal of the double similarity criterion and which applies between the first corresponding block CB t-1 in the search window SW t-1 and the reference block RB t in the image J t and which is noted: SC VS (Mk t , Mk t-1 + V)
- SC AT Mk t , Mk t + 1 + V
- SC VS ⁇ [I t (x k , y k ) - I t-1 (x k + V x , y k + V y )] 2
- SC AT ⁇ [I t (x k , y k ) - I t + 1 (x k + V x , y k - V y )] 2 (x k , y k ) ⁇ NG k
- the causal and anti-causal parts of the dual criterion as written above respectively represent the Euclidean distance between the block of reference and the candidate block in the causal image and in the anti-causal image, that is to say the sum of the squares of the differences in intensities of two pixels, one of which is in the reference block and the other of which is respectively in the candidate block of the causal image which corresponds to the vector V (t, t-1) and in the candidate block of the anti-causal image which corresponds to the vector V (t, t + 1).
- a vector V is defined by its two components V x and V y .
- V x and V y For each location x, y, there is a pixel in the reference block and a corresponding pixel in the candidate block. The intensity differences between the two pixels are calculated, are squared and the whole is summed. This represents a lot of calculation.
- SC C of the similarity criterion is calculated for each candidate block, and the block which gives the most convincing measure of similarity is chosen to determine the motion vector. So the number of differences squared which must be calculated to evaluate the similarity criterion is equal to the number of pixels that there are in the block: if it is a block of 8 x 8 pixels, it is a sum of 64 items.
- the pixels are taken two by two, at a rate of 1 in each block, separated by the same vector V.
- a vector V is defined for each candidate block.
- all the pixels are supposed to have the same motion vector. That is to say that, at the base of the FA-BMA process, it is a translational movement which connects two blocks, and that V x and V y do not vary for a given candidate block. If we define a block of 8 x 8 pixels in size, and if we define a SW search window which is more or less 10 or 15 pixels around the homologous block, there are 10 or 15 possible positions on each side of the block in x and in there. This means that the object can move more or less 10 or 15 pixels.
- the search window is 28 x 28 pixels, or 33 x 33 pixels. This means that, per candidate block, we must make 64 differences squared and summed for a single measure of similarity. The same is true for the estimation of the motion vector by means of the second anti-causal part SC A of the intensity similarity criterion.
- a specific exhaustive search policy noted F-SP, is used, according to which all the possible blocks of the search window are chosen, that is to say all the blocks which are referenced by their reference point, which is for example the central pixel, and by their dimensions. Each point in the search window is then a reference point for a candidate block.
- SC C the first causal criterion SC C
- the double criterion of similarity of intensity D-SC is therefore new in that it combines a causal part and an anti-causal part which are the same elementary similarity criterion applied at different times.
- D-SC SC VS [Mk t , Mk t-1 + V] + SC AT [Mk t , Mk t + 1 -V]
- the motion vector sought is that which minimizes the double criterion of similarity over the set of candidate blocks defined in the search windows SW t + 1 and SW t-1 in the images J t + 1 and J t-1 .
- the criterion of causal anti-causal similarity is a function or a combination of two causal and anti-causal parts of a similarity criterion intensity.
- L2 the Euclidean distance or norm
- L1 the sum of the absolute values of the differences in intensity.
- D-SC DL2 the Euclidean distance or norm
- L1 the sum of the absolute values of the differences in intensity.
- the present method comprises a sub-step 41 carried out in parallel with sub-step 42 in step 40, forming two chains 41, 42 ending in a TEST step 50 which applies in this case.
- V 0 be the vector determined by an elementary similarity criterion applied between the only reference and causal blocks
- V DLp be the vector determined by the exhaustive double causal anti-causal criterion.
- the vector VDLp is less good than the vector V 0 in the case of a non-linear non-continuous non-uniform movement since the vector V 0 is the best in the search window between the images J t and J t-1 . Due to this non-continuity the vector V DLp can be completely different from the true motion vector, from which it follows that the distance: Lp (V 0 ) ⁇ mx Lp (V DLp ) In formula (7) the factor m must be specified.
- the present method makes it possible to test the hypothesis of linearity, and, if it turns out that one is not in this hypothesis of linearity, it provides a solution which is locally acceptable.
- the use of the dual and adaptive criterion AD-SC DLp leads to quite remarkable results, all the more since it is very rare to fault the assumption of linearity between three temporal images.
- the double and adaptive criterion of intensity similarity denoted AD-SC DLp is based on the information of three time blocks and is constituted by a function of a part of causal criterion and part of an anti-causal criterion.
- This method makes it possible in particular to carry out a TEST on the causal part of this function and, at the end of this TEST, to deduce whether the linearity hypothesis is verified or not. This check is made on each block.
- the calculation on the causal part of the criterion is first performed providing the vector which minimizes the causal part of the criterion, which means that this TEST is carried out only once per reference block.
- This double and adaptive similarity criterion introduces a notion of temporal coherence, that is to say that the vector field that we find temporally verifies a linearity hypothesis, which imposes a constraint on the vector that one seeks.
- the application of this motion vector allows to filter according to the direction of movement. So the time filtering is applied between blocks that look alike instead of being done between blocks dissimilar because of the movement.
- the original image is decomposed in a step 101 by filtering it and by sub-sampling it.
- the neighborhood is 4 pixels around a pixel considered.
- the average is calculated on the quadruple of pixels. This produces a decomposed image at a first level denoted J ′ t .
- J ′ t a first level denoted
- this pyramid of images is produced on the image J t in step 101, on the causal image J t-1 , which gives J ' t-1 , J " t -1 in step 102 and on the anti-causal image J t + 1 , which gives J ' t + 1 , J " t + 1 in step 103.
- steps 110, 120, 130 the new method according to the invention which was noted above FA-BMA between the images to be processed and the causal and anti-causal images having the same resolution respectively.
- the number of calculations involved in searching for motion vectors is no longer proportional to the magnitude of the motion vector that we want to detect, but to this magnitude divided by two at the power n where n is the number of decomposition levels. So with two levels of decomposition compared to the normal resolution the number of computations is divided by four.
- step 110 a vector V " t of motion is determined by means of the three images at the coarsest resolution.
- step 120 a vector V ' t with intermediate resolution is sought: this is that is to say that we multiply by two the magnitude of the vector V " t found at the coarsest resolution and we determine the search windows in the intermediate causal and anti-causal images not around the location of the point of reference but around the end of this most probable vector determined using the coarsest resolution images.
- step 130 the operations in step 130 are repeated in the triplet of images at the finest resolution, which makes it possible to quickly determine a motion vector. V t at this finest resolution.
- a digital radiography device includes a source 1 of X-rays, a table 2 to receive the patient, a image intensifier device 3 coupled to a video tube 4, which provides data to a digital image processing system 5 comprising a processor. for implement the motion estimation method and noise filtering using this process.
- This processor includes several outputs including an output 6 coupled to a monitor 7 to view a sequence of radiographic images.
- the radiography device can work in fluoroscopic mode thanks to this system image processing.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Apparatus For Radiation Diagnosis (AREA)
- Picture Signal Circuits (AREA)
- Closed-Circuit Television Systems (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
Description
Claims (10)
- Procédé de traitement d'images pour l'estimation de mouvement dans une séquence d'images bruitées, incluant une étape de détermination d'un champ de vecteurs de mouvement d'une image temporelle à la suivante sur la base d'un algorithme de mise en correspondance de blocs, ce procédé comprenant des étapes de :acquisition d'une séquence de trois images temporelles successives incluant une image à traiter (Jt) entre une image antérieure appelée causale (Jt-1) et une image ultérieure appelée anti-causale (Jt+1),détermination d'un bloc de référence (RBt) dans l'image à traiter, et d'un couple de blocs localisés respectivement dans l'image causale et l'image anti-causale, (CBt-1,CBt+1) correspondant au mieux au bloc de référence (RBt) selon un critère de similarité d'intensité, et détermination d'un vecteur de mouvement relatif au bloc de référence sur la base de la localisation des blocs de ce couple vis-à-vis de la localisation du bloc de référence.
- Procédé selon la revendication 1, où le critère de similarité d'intensité est basé sur une hypothèse de mouvement linéaire, continu et uniforme du bloc correspondant dans l'image causale au bloc correspondant dans l'image anti-causale, pour sélectionner des blocs correspondants symétriques par rapport au bloc de référence et pour déterminer des vecteurs de mouvement symétriques du bloc de référence vers chacun des blocs correspondants.
- Procédé selon l'une des revendications 1 ou 2, où le critère de similarité d'intensité appelé critère double est une fonction d'un premier critère de similarité élémentaire appelé partie causale qui relie le bloc de référence au bloc correspondant dans l'image causale et d'un second critère de similarité élémentaire appelé partie anti-causale qui relie le bloc de référence au bloc correspondant dans l'image anti-causale, ces parties causale et anti-causale étant identiques et appliquées à des instants différents relatifs aux images causale et anti-causales.
- Procédé selon la revendication 3, où la fonction pour former le critère double à partir des parties causale et anti-causale est choisie parmi une somme, une moyenne et une combinaison linéaire.
- Procédé selon l'une des revendications 3 ou 4, où l'étape de détermination des blocs correspondants comprend un test appliqué sur le mouvement entre le bloc correspondant dans l'image causale et le bloc correspondant dans l'image anti-causale pour déterminer si une hypothèse que ce mouvement est linéaire, continu et uniforme est vérifiée ou non, et en fonction de ce test, une décision est prise que:si cette hypothèse est vérifiée le vecteur de mouvement est déterminé par le critère double de similarité d'intensité,et si cette hypothèse n'est pas vérifiée le vecteur de mouvement est déterminé par un critère élémentaire de similarité d'intensité entre le bloc de référence et le bloc correspondant anti-causal,
- Procédé selon l'une des revendications 3 à 5, où les critères élémentaires de similarité d'intensité sont des distances d'un type appelé Minkowski.
- Procédé de traitement d'image selon l'une des revendications 1 à 6 comprenant:une étape de formation de trois pyramides d'images à différentes résolutions à partir de l'image à traiter, à partir de l'image causale et à partir de l'image anti-causale,une étape de détermination d'un vecteur de mouvement par l'application du critère de similarité sur les trois images de plus faible résolution dérivées de l'image à traiter, de l'image causale et de l'image anticausale,une étape de report du vecteur mouvement trouvé à la plus faible résolution dans les images dérivées de l'image à traiter, de l'image causale et de l'image anticausale à la résolution immédiatement supérieure et de recherche d'un nouveau vecteur de mouvement autour du vecteur reporté à cette résolution immédiatement supérieure,une répétition de cette étape jusqu'à la résolution la meilleure pour trouver le vecteur de mouvement recherché.
- Procédé de traitement d'image pour la réduction du bruit dans une séquence d'images comprenant des étapes selon l'une des revendications 1 à 7 pour l'évaluation de vecteurs de mouvements entre les images de la séquence, et des étapes de filtrage temporel récursif du bruit avec compensation de mouvement au moyen de ces vecteurs.
- Appareil d'imagerie médicale qui comprend un système d'acquisition d'une séquence d'images, et un système de traitement d'image qui a accès aux données des images et à un système d'affichage des images de la séquence et qui comprend un processeur mettant en oeuvre un procédé de traitement d'image selon l'une des revendications 1 à 8.
- Appareil à rayons X qui comprend un système d'acquisition d'une séquence d'images, et un système de traitement d'image qui a accès aux données des images et à un système d'affichage des images de la séquence et qui comprend un processeur mettant en oeuvre un procédé de traitement d'image selon l'une des revendications 1 à 8.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
FR9801906 | 1998-02-17 | ||
FR9801906 | 1998-02-17 |
Publications (2)
Publication Number | Publication Date |
---|---|
EP0936575A1 true EP0936575A1 (fr) | 1999-08-18 |
EP0936575B1 EP0936575B1 (fr) | 2004-07-28 |
Family
ID=9523058
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
EP99200360A Expired - Lifetime EP0936575B1 (fr) | 1998-02-17 | 1999-02-08 | Procédé de traitement d'images pour l'estimation de mouvement dans une séquence d'images |
Country Status (4)
Country | Link |
---|---|
US (1) | US6154519A (fr) |
EP (1) | EP0936575B1 (fr) |
JP (1) | JPH11331644A (fr) |
DE (1) | DE69918877T2 (fr) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8428373B2 (en) | 2002-07-18 | 2013-04-23 | Lg Electronics Inc. | Apparatus for determining motion vectors and a reference picture index for a current block in a picture to be decoded |
US20210033440A1 (en) * | 2019-07-29 | 2021-02-04 | Supersonic Imagine | Ultrasonic system for detecting fluid flow in an environment |
Families Citing this family (24)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0926626A1 (fr) * | 1997-12-23 | 1999-06-30 | Koninklijke Philips Electronics N.V. | Procédé de traitement d'image pour réduire le bruit dans une image d'une séquence de trois images temporelles et appareil d'imagerie médicale mettant en oeuvre un tel procédé |
KR100840133B1 (ko) * | 2000-05-18 | 2008-06-23 | 코닌클리케 필립스 일렉트로닉스 엔.브이. | Mc 업 컨버젼에서 헤일로를 감소시키는 움직임 추정기 |
US6934332B1 (en) | 2001-04-24 | 2005-08-23 | Vweb Corporation | Motion estimation using predetermined pixel patterns and subpatterns |
US6891890B1 (en) * | 2001-04-24 | 2005-05-10 | Vweb Corporation | Multi-phase motion estimation system and method |
US6813315B1 (en) | 2001-04-24 | 2004-11-02 | Vweb Corporation | Motion estimation using multiple search windows |
US7233696B2 (en) * | 2002-07-12 | 2007-06-19 | Hill Richard K | Apparatus and method for characterizing digital images using a two axis image sorting technique |
US7620221B2 (en) * | 2003-01-29 | 2009-11-17 | Koninklijke Philips Electronics N.V. | System and method for enhancing an object of interest in noisy medical images |
US7440628B2 (en) * | 2004-08-31 | 2008-10-21 | Siemens Medical Solutions Usa, Inc. | Method and system for motion correction in a sequence of images |
US20060239344A1 (en) * | 2005-04-25 | 2006-10-26 | Ashish Koul | Method and system for rate control in a video encoder |
JP4915836B2 (ja) * | 2005-12-15 | 2012-04-11 | 株式会社日立メディコ | 画像処理装置の作動方法、プログラム、及び装置 |
US7830565B2 (en) * | 2006-06-27 | 2010-11-09 | Motorola, Inc. | Image capture device with rolling band shutter |
FR2924254B1 (fr) * | 2007-11-23 | 2010-01-01 | Gen Electric | Procede de traitement d'images en radioscopie interventionnelle |
US8223235B2 (en) * | 2007-12-13 | 2012-07-17 | Motorola Mobility, Inc. | Digital imager with dual rolling shutters |
JP5053982B2 (ja) * | 2008-12-05 | 2012-10-24 | 株式会社東芝 | X線診断装置および画像処理装置 |
US8189080B2 (en) * | 2009-04-03 | 2012-05-29 | Sony Corporation | Orientation-based approach for forming a demosaiced image, and for color correcting and zooming the demosaiced image |
US8233747B2 (en) * | 2009-04-03 | 2012-07-31 | Sony Corporation | Method and apparatus for forming super resolution images from raw data representative of color filter array images |
CN102549621B (zh) * | 2009-10-06 | 2016-08-31 | 皇家飞利浦电子股份有限公司 | 用于处理至少包含代表生物中的周期现象的分量的信号的方法和系统 |
JP5586209B2 (ja) * | 2009-11-04 | 2014-09-10 | 三菱電機株式会社 | 画像信号処理装置 |
WO2011068783A1 (fr) * | 2009-12-03 | 2011-06-09 | The United States Of America, As Represented By The Secretary, Department Of Health And Human Services | Applications d'imagerie à rapport signal sur bruit amélioré à l'aide d'une série chronologique d'images |
EP2550641B1 (fr) * | 2010-03-24 | 2017-10-18 | Koninklijke Philips N.V. | Système et procédé permettant de produire une image d'un objet physique |
KR102274320B1 (ko) | 2014-10-16 | 2021-07-07 | 삼성전자주식회사 | 영상 처리 방법 및 장치 |
US10332243B2 (en) * | 2016-12-12 | 2019-06-25 | International Business Machines Corporation | Tampering detection for digital images |
US10523961B2 (en) | 2017-08-03 | 2019-12-31 | Samsung Electronics Co., Ltd. | Motion estimation method and apparatus for plurality of frames |
US11154251B2 (en) * | 2018-02-10 | 2021-10-26 | The Governing Council Of The University Of Toronto | System and method for classifying time series data for state identification |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5337231A (en) * | 1992-03-31 | 1994-08-09 | General Electric Company | View to view image correction for object motion with truncated data |
WO1995007591A1 (fr) * | 1993-09-08 | 1995-03-16 | Thomson Multimedia S.A. | Procede et appareil d'evaluation de mouvement au moyen de l'assortiment des blocs |
EP0780795A1 (fr) * | 1995-12-22 | 1997-06-25 | THOMSON multimedia | Procédé d'estimation de mouvement |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5473384A (en) * | 1993-12-16 | 1995-12-05 | At&T Corp. | Method of and system for enhancing distorted graphical information |
FR2736181A1 (fr) * | 1995-06-30 | 1997-01-03 | Philips Electronique Lab | Procede de traitement d'images pour la reduction du bruit dans une image d'une sequence d'images numeriques et dispositif mettant en oeuvre ce procede |
FR2736182A1 (fr) * | 1995-06-30 | 1997-01-03 | Philips Electronique Lab | Procede de traitement d'images pour la reduction du bruit dans une image d'une sequence d'images numeriques et dispositif mettant en oeuvre ce procede |
US5717463A (en) * | 1995-07-24 | 1998-02-10 | Motorola, Inc. | Method and system for estimating motion within a video sequence |
US5790686A (en) * | 1995-09-19 | 1998-08-04 | University Of Maryland At College Park | DCT-based motion estimation method |
-
1999
- 1999-02-08 EP EP99200360A patent/EP0936575B1/fr not_active Expired - Lifetime
- 1999-02-08 DE DE69918877T patent/DE69918877T2/de not_active Expired - Fee Related
- 1999-02-12 US US09/249,967 patent/US6154519A/en not_active Expired - Fee Related
- 1999-02-16 JP JP11037497A patent/JPH11331644A/ja not_active Withdrawn
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5337231A (en) * | 1992-03-31 | 1994-08-09 | General Electric Company | View to view image correction for object motion with truncated data |
WO1995007591A1 (fr) * | 1993-09-08 | 1995-03-16 | Thomson Multimedia S.A. | Procede et appareil d'evaluation de mouvement au moyen de l'assortiment des blocs |
EP0780795A1 (fr) * | 1995-12-22 | 1997-06-25 | THOMSON multimedia | Procédé d'estimation de mouvement |
Non-Patent Citations (3)
Title |
---|
KERDRANVAT M: "HIERARCHICAL MOTION ESTIMATION AND MOTION INFORMATION ENCODING", SIGNAL PROCESSING OF HDTV, 2, TURIN, AUG. 30 - SEPT. 1, 1989, no. 3, 30 August 1989 (1989-08-30), CHIARIGLIONE L, pages 101 - 108, XP000215230 * |
NEMOUCHI Y ET AL: "A REFERENCE DATA ORIENTED APPROACH FOR IMPLEMENTING A MULTIRESOLUTION BLOCK-MATCHING ALGORITHM", INTERNAT. SYMP. ON CIRCUITS AND SYSTEMS (ISCAS), ATLANTA, MAY 12 - 15, 1996, vol. 4, 12 May 1996 (1996-05-12), IEEE, pages 249 - 252, XP000704583 * |
SANG-YEON KIM ET AL: "A SYSTOLIC REALIZATION OF SUMMETRIC BLOCK MATCHING ALGORITHM FOR HD-MAC SYSTEM", IEEE TRANSACTIONS ON CONSUMER ELECTRONICS, vol. 39, no. 3, 1 August 1993 (1993-08-01), pages 277 - 283, XP000396292 * |
Cited By (27)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8428373B2 (en) | 2002-07-18 | 2013-04-23 | Lg Electronics Inc. | Apparatus for determining motion vectors and a reference picture index for a current block in a picture to be decoded |
US8463058B2 (en) | 2002-07-18 | 2013-06-11 | Lg Electronics Inc. | Calculation method for prediction motion vector |
US8467621B2 (en) | 2002-07-18 | 2013-06-18 | Lg Electronics Inc. | Method of determining motion vectors and a reference picture index for a current block in a picture to be decoded |
US8467622B2 (en) | 2002-07-18 | 2013-06-18 | Lg Electronics Inc. | Method of determining motion vectors and a reference picture index for a current block in a picture to be decoded |
US8467620B2 (en) | 2002-07-18 | 2013-06-18 | Lg Electronics Inc. | Method of determining motion vectors and a reference picture index for a current block in a picture to be decoded |
US8472738B2 (en) | 2002-07-18 | 2013-06-25 | Lg Electronics Inc. | Apparatus for determining motion vectors and a reference picture index for a current block in a picture to be decoded |
US8509550B2 (en) | 2002-07-18 | 2013-08-13 | Lg Electronics Inc. | Apparatus for determining motion vectors and a reference picture index for a current block in a picture to be decoded |
US8548264B2 (en) | 2002-07-18 | 2013-10-01 | Lg Electronics Inc. | Apparatus for predicting a motion vector for a current block in a picture to be decoded |
US8565544B2 (en) | 2002-07-18 | 2013-10-22 | Lg Electronics Inc. | Apparatus for predicting a motion vector for a current block in a picture to be decoded |
US8571335B2 (en) | 2002-07-18 | 2013-10-29 | Lg Electronics Inc. | Calculation method for prediction motion vector |
US8634666B2 (en) | 2002-07-18 | 2014-01-21 | Lg Electronics Inc. | Apparatus for determining motion vectors and a reference picture index for a current block in a picture to be decoded |
US8634667B2 (en) | 2002-07-18 | 2014-01-21 | Lg Electronics Inc. | Method of predicting a motion vector for a current block in a current picture |
US8639048B2 (en) | 2002-07-18 | 2014-01-28 | Lg Electronics Inc. | Method of determining motion vectors and a reference picture index for a current block in a picture to be decoded |
US8644631B2 (en) | 2002-07-18 | 2014-02-04 | Lg Electronics Inc. | Method of predicting a motion vector for a current block in a current picture |
US8644630B2 (en) | 2002-07-18 | 2014-02-04 | Lg Electronics Inc. | Apparatus for determining motion vectors and a reference picture index for a current block in a picture to be decoded |
US8649621B2 (en) | 2002-07-18 | 2014-02-11 | Lg Electronics Inc. | Apparatus for determining motion vectors and a reference picture index for a current block in a picture to be decoded |
US8649622B2 (en) | 2002-07-18 | 2014-02-11 | Lg Electronics Inc. | Method of determining motion vectors and a reference picture index for a current block in a picture to be decoded |
US8655089B2 (en) | 2002-07-18 | 2014-02-18 | Lg Electronics Inc. | Apparatus for determining motion vectors and a reference picture index for a current block in a picture to be decoded |
US8712172B2 (en) | 2002-07-18 | 2014-04-29 | Lg Electronics Inc. | Method of predicting a motion vector for a current block in a current picture |
US8908983B2 (en) | 2002-07-18 | 2014-12-09 | Lg Electronics Inc. | Method of predicting a motion vector for a current block in a current picture |
US9544589B2 (en) | 2002-07-18 | 2017-01-10 | Lg Electronics Inc. | Method of predicting a motion vector for a current block in a current picture |
US9544591B2 (en) | 2002-07-18 | 2017-01-10 | Lg Electronics Inc. | Method of predicting a motion vector for a current block in a current picture |
US9544590B2 (en) | 2002-07-18 | 2017-01-10 | Lg Electronics Inc. | Method of predicing a motion vector for a current block in a current picture |
US9560354B2 (en) | 2002-07-18 | 2017-01-31 | Lg Electronics Inc. | Method of predicting a motion vector for a current block in a current picture |
US10425639B2 (en) | 2002-07-18 | 2019-09-24 | Lg Electronics Inc. | Method of predicting a motion vector for a current block in a current picture |
US10897613B2 (en) | 2002-07-18 | 2021-01-19 | Lg Electronics Inc. | Method of predicting a motion vector for a current block in a current picture |
US20210033440A1 (en) * | 2019-07-29 | 2021-02-04 | Supersonic Imagine | Ultrasonic system for detecting fluid flow in an environment |
Also Published As
Publication number | Publication date |
---|---|
JPH11331644A (ja) | 1999-11-30 |
US6154519A (en) | 2000-11-28 |
DE69918877T2 (de) | 2005-07-28 |
EP0936575B1 (fr) | 2004-07-28 |
DE69918877D1 (de) | 2004-09-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP0936575B1 (fr) | Procédé de traitement d'images pour l'estimation de mouvement dans une séquence d'images | |
EP0751483B1 (fr) | Procédé de filtrage temporel du bruit dans une image d'une séquence d'images numériques et dispositif mettant en oeuvre ce procédé | |
EP2489007B1 (fr) | Correction de flou d'image à l'aide d'une image spatiale antérieure | |
EP0751482B1 (fr) | Procédé de filtrage temporel du bruit dans une image d'une séquence d'images numériques et dispositif mettant en oeuvre ce procédé | |
US4503461A (en) | Multiple measurement noise reducing system using space-variant filters | |
US7602989B2 (en) | Realtime 2D deconvolution system and method | |
EP0871143B1 (fr) | Système de traitement d'une séquence d'images bruitées, et appareil d'examen médical incluant un tel système | |
EP2164040B1 (fr) | Système et procédé pour une grande qualité d'image et l'interpolation vidéo | |
FR2831013A1 (fr) | Procede et systeme pour traiter une image fluoroscopique | |
FR2766946A1 (fr) | Procede et dispositif de pretraitement pour estimation de mouvement | |
KR102106537B1 (ko) | 하이 다이나믹 레인지 영상 생성 방법 및, 그에 따른 장치, 그에 따른 시스템 | |
WO2014023904A1 (fr) | Procede et dispositif pour la reconstruction d'images a super-resolution | |
EP0926626A1 (fr) | Procédé de traitement d'image pour réduire le bruit dans une image d'une séquence de trois images temporelles et appareil d'imagerie médicale mettant en oeuvre un tel procédé | |
FR2666426A1 (fr) | Procede de correction des mesures de densite optique effectuees sur un film radiographique. | |
FR2564627A1 (fr) | Dispositif d'application de fenetre dans un systeme de visualisation | |
JP5809920B2 (ja) | 画像処理方法及び装置 | |
US8675080B2 (en) | Motion estimation in imaging systems | |
FR2790852A1 (fr) | Procede d'evaluation d'images codees, dispositif mettant en oeuvre le procede et utilisation du dispositif et procede | |
EP0926625B1 (fr) | Procédé de traitement d'image incluant des étapes de filtrage spatio-temporel du bruit et appareil d'imagerie médicale mettant en oeuvre ce procédé | |
EP0821322B1 (fr) | Procédé de filtrage temporel du bruit dans une image d'une séquence d'images numérisées et dispositif mettant en oeuvre ce procédé | |
CN105593899A (zh) | 用于在图像序列中的噪声降低的方法 | |
JPH06507992A (ja) | 画像平滑化のための装置及び方法 | |
KR102007601B1 (ko) | 비디오 잡음 제거를 위한 효율적인 패치 기반 방법 | |
Daniel et al. | Retinal image enhancement using wavelet domain edge filtering and scaling | |
JPH1074192A (ja) | 画像処理方法およびその装置 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): DE FR GB NL |
|
AX | Request for extension of the european patent |
Free format text: AL;LT;LV;MK;RO;SI |
|
17P | Request for examination filed |
Effective date: 20000218 |
|
AKX | Designation fees paid |
Free format text: DE FR GB NL |
|
17Q | First examination report despatched |
Effective date: 20030613 |
|
GRAP | Despatch of communication of intention to grant a patent |
Free format text: ORIGINAL CODE: EPIDOSNIGR1 |
|
RTI1 | Title (correction) |
Free format text: IMAGE PROCESSING METHOD FOR MOTION ESTIMATION IN A SEQUENCE OF IMAGES |
|
GRAS | Grant fee paid |
Free format text: ORIGINAL CODE: EPIDOSNIGR3 |
|
GRAA | (expected) grant |
Free format text: ORIGINAL CODE: 0009210 |
|
AK | Designated contracting states |
Kind code of ref document: B1 Designated state(s): DE FR GB NL |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: NL Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20040728 Ref country code: GB Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20040728 |
|
REG | Reference to a national code |
Ref country code: GB Ref legal event code: FG4D Free format text: NOT ENGLISH |
|
REF | Corresponds to: |
Ref document number: 69918877 Country of ref document: DE Date of ref document: 20040902 Kind code of ref document: P |
|
NLV1 | Nl: lapsed or annulled due to failure to fulfill the requirements of art. 29p and 29m of the patents act | ||
GBV | Gb: ep patent (uk) treated as always having been void in accordance with gb section 77(7)/1977 [no translation filed] |
Effective date: 20040728 |
|
PLBE | No opposition filed within time limit |
Free format text: ORIGINAL CODE: 0009261 |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: NO OPPOSITION FILED WITHIN TIME LIMIT |
|
26N | No opposition filed |
Effective date: 20050429 |
|
PGFP | Annual fee paid to national office [announced via postgrant information from national office to epo] |
Ref country code: DE Payment date: 20070410 Year of fee payment: 9 |
|
PGFP | Annual fee paid to national office [announced via postgrant information from national office to epo] |
Ref country code: FR Payment date: 20070226 Year of fee payment: 9 |
|
REG | Reference to a national code |
Ref country code: FR Ref legal event code: ST Effective date: 20081031 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: DE Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20080902 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: FR Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20080229 |